Home > Research > Publications & Outputs > An Online Incremental Learning Approach for Con...

Research

Associated organisational units

Electronic data

  • m2033

    Accepted author manuscript, 1.44 MB, PDF document

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

View graph of relations

An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Forthcoming

Standard

An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms. / Alsomali, Mohammad; Rodrigues-Filho, Roberto ; Soriano Marcolino, Leandro et al.
ECAI: European Conference On Artificial Intelligence. 2024. 2033.

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Harvard

Alsomali, M, Rodrigues-Filho, R, Soriano Marcolino, L & Porter, B 2024, An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms. in ECAI: European Conference On Artificial Intelligence., 2033.

APA

Alsomali, M., Rodrigues-Filho, R., Soriano Marcolino, L., & Porter, B. (in press). An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms. In ECAI: European Conference On Artificial Intelligence Article 2033

Vancouver

Alsomali M, Rodrigues-Filho R, Soriano Marcolino L, Porter B. An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms. In ECAI: European Conference On Artificial Intelligence. 2024. 2033

Author

Alsomali, Mohammad ; Rodrigues-Filho, Roberto ; Soriano Marcolino, Leandro et al. / An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms. ECAI: European Conference On Artificial Intelligence. 2024.

Bibtex

@inproceedings{4d8cb554fdec43b38bc2d415c934c3fd,
title = "An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms",
abstract = "This paper introduces Dynamic Bayesian Optimisationfor Multi-Arm Bandits (DBO-MAB), an algorithm that dynamicallyadapts hyperparameters of multi-arm bandit algorithms using incrementalBayesian optimisation. DBO-MAB addresses the challengeof tuning hyperparameters in uncertain and dynamic environments,particularly for applications like web server optimisation. It uses adynamic range adjustment approach based on the interquartile mean(IQM) of observed rewards to focus the search space on promisingregions. Evaluated across diverse static and dynamic environments,DBO-MAB outperforms state-of-the-art algorithms such as BootstrappedUCB and f-Discounted-Sliding-Window Thompson Sampling,reducing average response time by ≈ 55%.",
author = "Mohammad Alsomali and Roberto Rodrigues-Filho and {Soriano Marcolino}, Leandro and Barry Porter",
year = "2024",
month = jul,
day = "4",
language = "English",
booktitle = "ECAI",

}

RIS

TY - GEN

T1 - An Online Incremental Learning Approach for Configuring Multi-arm Bandits Algorithms

AU - Alsomali, Mohammad

AU - Rodrigues-Filho, Roberto

AU - Soriano Marcolino, Leandro

AU - Porter, Barry

PY - 2024/7/4

Y1 - 2024/7/4

N2 - This paper introduces Dynamic Bayesian Optimisationfor Multi-Arm Bandits (DBO-MAB), an algorithm that dynamicallyadapts hyperparameters of multi-arm bandit algorithms using incrementalBayesian optimisation. DBO-MAB addresses the challengeof tuning hyperparameters in uncertain and dynamic environments,particularly for applications like web server optimisation. It uses adynamic range adjustment approach based on the interquartile mean(IQM) of observed rewards to focus the search space on promisingregions. Evaluated across diverse static and dynamic environments,DBO-MAB outperforms state-of-the-art algorithms such as BootstrappedUCB and f-Discounted-Sliding-Window Thompson Sampling,reducing average response time by ≈ 55%.

AB - This paper introduces Dynamic Bayesian Optimisationfor Multi-Arm Bandits (DBO-MAB), an algorithm that dynamicallyadapts hyperparameters of multi-arm bandit algorithms using incrementalBayesian optimisation. DBO-MAB addresses the challengeof tuning hyperparameters in uncertain and dynamic environments,particularly for applications like web server optimisation. It uses adynamic range adjustment approach based on the interquartile mean(IQM) of observed rewards to focus the search space on promisingregions. Evaluated across diverse static and dynamic environments,DBO-MAB outperforms state-of-the-art algorithms such as BootstrappedUCB and f-Discounted-Sliding-Window Thompson Sampling,reducing average response time by ≈ 55%.

M3 - Conference contribution/Paper

BT - ECAI

ER -